rNexa
rNexa docs.
Everything on this page is also inside the app: open Help in the top bar and search it there while you work.
Help · top centre of the workspace, first in the panel row.
01Getting started
Three steps from zero to building.
- Add a project — open any folder on this machine, create a fresh workspace, or add a server to work on it remotely.
- Start a lane on it — chats are AI agents that build for you, shells are real terminals.
- Stay in control — everything runs on your machine, on your own provider accounts.
02Chats, shells and agents
Every lane is one worker. + New opens a chat (rNexa, Claude, Codex), a shell, or the Agent Orchestrator that runs a team of them.
A shell can take a role — Project Lead, Builder, Frontend, Backend, Auditor — so it works to that brief. Lanes run at the same time and keep going while you look somewhere else.
03Queue and steer
Keep typing while the agent works. Send during a task and the message waits in the follow-up queue, in order, for the moment the agent frees up.
Steer sends it right now instead — the running task keeps going and takes the new direction with it. Stop ends the current task without closing the thread.
04Threads and forks
Every conversation is saved and resumable. Close a thread and it stays in the panel under its project, ready to pick up where you left off.
Fork a thread to branch an experiment without losing the original — the original becomes the mother thread and lists its forks. The coloured dot shows live status: green working, yellow waiting for you.
05Focus and Power view
The two icons beside your machine name switch the layout: Focus shows one foreground lane, Power shows up to six at once.
Hidden lanes keep running either way — switching view never stops an agent.
06Agent mode
Let a chat drive rNexa itself. With Agent mode on, that chat can open panels, switch projects, move files and use the browser for you.
It is per chat, and you can turn it off mid-conversation — the agent keeps working, it just stops touching the app.
07Computer use
Your desktop, when you allow it. rNexa Computer lets an agent see the screen and use apps outside rNexa, with a live view so you watch every action.
You set the permission: ask before each action, run automatically, or full access for a short session. Sensitive screens stay hidden from capture.
08rNexa Browser
A real browser inside the workspace. Open sites and your running app beside your work, or hand the browser to an agent that clicks, types and reports what it sees.
It keeps its own profile, so your normal signed-in browser is untouched.
09rNexa Design
Design to a real direction, not a template. Turn it on in a chat and every design job follows the rNexa Design playbook: research first, then real options to choose from.
Made for a page, screen, dashboard or a redesign — the chip in the composer shows it is on.
10Commander
Talk to rNexa, hands free. Commander is the voice lane: speak a task and it runs it, reports back, and can steer the agents already working.
It answers out loud and keeps a caption of everything it heard and did.
11Plan and Pad
Plan collects your tasks and the ones your agents are working through, across all projects. Agents tick off their own steps, so the board shows real progress.
Pad is a drawing canvas and scratch space — sketch a layout, then hand the sketch to a chat. It remembers its size and position.
12Review what an agent did
Nothing lands unseen. The task card opens the full run: steps, tools, approvals, diffs and the result.
Change Trail lists changed files with restore points, so you can undo one task without undoing your day.
13Slash commands and @ mentions
Type / for actions the chat runs on itself, and @ to pull in a project, file, plugin or skill. The menu offers only what the current provider actually supports.
14Projects, servers and parking
Load as many projects as your machine handles. Add a server to work on a remote machine with the same lanes, files and preview you use locally.
Quiet projects park automatically when room is needed and wake instantly when you open them — nothing is lost.
15Providers and accounts
Your own accounts, your own limits. Connect Claude, Codex and the rest once in Settings — rNexa uses their sign-in on this machine.
Add more than one account per provider to see usage per account and keep working when one is spent.
16Plugins, MCP and keys
Open the Plugins tab and add what you use — GitHub, Supabase, Figma, Slack and more. Authorization happens only when a plugin needs it.
Power users can add their own MCP server under + Add your own. Unreviewed — only add servers you trust.
Keys and secrets
Keys live in Settings → Secrets, encrypted on this machine, and your tools can use them without them ever being shown again. A database key is the connection string your own database gives you — rNexa never sees it in chat.
17Skills
A skill is a reusable playbook your agents follow — a release checklist, a review pass, your team's rules.
Open the Skill Center to build one with your agent, import one, or start from the gallery.
18Appearance and theme sharing
Tune surface, accent, darkness, tint, blur, background and dim in Appearance — or switch to Solid.
Copy your whole theme as a short code and share it; pasting one applies the look instantly.
19Keyboard shortcuts
The full list lives in Settings → Shortcuts. Voice dictation is available in chat composers on supported systems.
20License and devices
One license covers up to 2 machines.
See this device in Settings → General, and free a slot from your account page on the web.
21Still stuck?
Send it from inside the app — Settings → Feedback reaches us with the build you are on. Or email [email protected].
Never send API keys, tokens or passwords in a message. We never ask for them.